The importance of objectively valid data is well established in clinical medicine. Such data include an accurate recording of a patient’s clinical history; evaluation of signs and symptoms of illness; and measurement of various routine indicators, such as granulocyte and platelet counts, serum glucose, electrolytes, and liver function tests.
Maurie Markman, MD
The importance of objectively valid data is well established in clinical medicine. Such data include an accurate recording of a patient’s clinical history; evaluation of signs and symptoms of illness; and measurement of various routine indicators, such as granulocyte and platelet counts, serum glucose, electrolytes, and liver function tests, and more specialized biomarkers, such as blood cultures and serum antibody levels to potential infectious pathogens. They also include imaging studies that hopefully help inform the differential diagnosis of a new medical condition or the status of an existing ailment.
In oncology, one must add to these general data elements those more specific to the cancer itself, including pathology, tumor staging, unique radiographic studies, and biological markers of the presence, extent, and progress of the cancer. Specific information related to the impact of treatment on the individual patient through positive signals such as a decrease in tumor markers and negative indicators such as bone marrow suppression must also be added. This description of the relevance of data in cancer management is understood by all who care for patients with cancer.
However, over time there have been rather dramatic, even revolutionary, changes in the types and complexity of clinically relevant data, the methods of recording what is collected (eg, electronic medical record), and the truly massive volume of information available which can positively or negatively influence the quality of care provided and the ultimate outcomes.
For example, one only needs to briefly consider the large number of new antineoplastic agents the FDA has approved for use in individual malignancies (eg, breast, non–small cell lung, colon, renal cell cancers, etc) compared with a mere 10 years ago to appreciate the growing volume of knowledge that oncologists must master to optimally care for patients in a specific clinical setting. The need for robust, easy-to-use, and up-to-date decision support tools is an increasingly critical requirement within the oncology community.
However, the goal of this commentary is to highlight another critical aspect of the rapid changes in both the quantity of data available and the innovative technology increasingly employed to collect this information. These developments have the potential to truly revolutionize cancer care through the understanding and meaningful use of what has been labeled real-world data.
Consider this hypothetical but certainly very real-world scenario: A 73-year-old modestly obese woman with a more than 25-year history of well-controlled insulin- dependent diabetes who has suffered a silent myocardial infarction 4 months before a diagnosis of metastatic triple negative breast cancer seeks your recommendation for therapy. Although the patient has documented disease in the lung and bone, her ECOG performance status is 1, and she has no evident cardiac symptoms.
A novel antineoplastic agent was approved for clinical use 1 year ago in this theoretical setting, with the drug shown to substantially improve both progression-free and overall survival rates compared with those of a standard-of-care control arm. Based on published phase 3 trial results, it is difficult to assess clearly the safety of this drug for this individual patient because, not surprisingly, few research participants in this landmark study were: (a) more than 70 years old; (b) modestly obese; (c) had a prolonged history of insulin-dependent diabetes; or (d) had a recent myocardial insult.
However, what if one could easily search a large population database potentially involving thousands of patients treated outside of clinical trial settings since the introduction of this agent to find a group of individuals with similar baseline characteristics such as age, weight, and comorbidities such as diabetes and recent myocardial injury? Evidence that the drug had been safely delivered in similar patients whose data were available in this database would be reassuring to the physician, the patient, and her family, indicating that using this agent could be a viable option despite the lack of clinical trial data addressing this critical point. Conversely, evidence that serious adverse events were associated with the drug’s administration (eg, hospitalization for complications of diabetes, new cardiac event) would likely and appropriately be an argument against its use.
It is not difficult to see the potential for this strategy; it could help inform clinical decisions in common settings where the current standard of care or novel single agents or combinations are employed in multiple patient populations that are poorly represented in the clinical trials leading to regulatory approval or in subsequently published studies that extended the use of the specific regimens in question.
Another possible use of such real-world databases would be updating information regarding potential adverse events associated with anti-neoplastic strategies appropriately approved for clinical use based on demonstrated efficacy, but with a limited number of patients treated when initial regulatory permission was granted.
For example, TRK inhibitors have been shown to be highly efficacious in a rare group of adult and pediatric cancers with demonstrated NTRK or ROS1 fusion abnormalities. As a result of the quite modest incidence of these cancers, the full spectrum of possible adverse effects, and their severity, was not fully characterized at the time of the initial drug approval. In a recent report, investigators retrospectively reviewed the clinical course of 96 patients treated with this class of agents, which has helped to more fully characterize the risks associated with this therapy, including specific, unique neurological toxicities (eg, withdrawal pain).
Finally, the potential utility of real-world data relates to the precision medicine arena. As more patients undergo advanced genomic testing in the search for unique cancer-specific molecular abnormalities that may be targeted, it is highly likely that oncologists will attempt to treat individual patients “off-label” with a variety of approaches based on existing and rapidly evolving medical knowledge. Hopefully, they would be in a better position to know whether their own patient might be an appropriate candidate to receive a specific therapeutic regimen if these “N-of-1” experiences involving evidence for or against clinical benefit (eg, tumor shrinkage or declines in cancer biomarkers, improvement in cancer-related symptoms, or extended time to subsequent disease progression) in specific clinical settings, such as those with a molecular abnormality detected in a particular tumor type, were combined and made widely available.
Future efforts that ensure essential patient security and privacy will transform rather stagnant data from electronic medical records into vitally relevant information that will help optimize the quality of care and clinical outcomes for the next patient we see in our office.
Liu D, Flory J, Lin A, et al. Characterization of on-target adverse events caused by TRK inhibitor therapy. Ann Oncol. 2020;31(9):1207-1215. doi:10.1016/j.annonc.2020.05.006